Sound storage devices may detect a level of a signal input from an instrument to automatically start recording when the detected level of the signal is higher than a predetermined threshold value. However, these sound storage devices cannot always start recording at an intended timing, for example, because the tones from playing instruments vary widely from player to player, and the levels of signals output from the instruments also vary widely. For example, the level of a signal of a large sound such as a note, chord, or passage played fortissimo is high, while the level of a signal of pianissimo with delicate expression is low. Therefore, if the start of the automatic recording is determined based only on the level of the signal, it is difficult for the storage device to determine the start of the automatic recording when a low-level signal based on delicate expression is input. As a result, the musical performance may not be recorded.